US2160603A - Comparometer - Google Patents

Comparometer Download PDF

Info

Publication number
US2160603A
US2160603A US161916A US16191637A US2160603A US 2160603 A US2160603 A US 2160603A US 161916 A US161916 A US 161916A US 16191637 A US16191637 A US 16191637A US 2160603 A US2160603 A US 2160603A
Authority
US
United States
Prior art keywords
card
orifices
profile
rails
orifice
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Lifetime
Application number
US161916A
Inventor
Richard G Sagebeer
Original Assignee
Richard G Sagebeer
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Richard G Sagebeer filed Critical Richard G Sagebeer
Priority to US161916A priority Critical patent/US2160603A/en
Application granted granted Critical
Publication of US2160603A publication Critical patent/US2160603A/en
Anticipated expiration legal-status Critical
Expired - Lifetime legal-status Critical Current

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06KRECOGNITION OF DATA; PRESENTATION OF DATA; RECORD CARRIERS; HANDLING RECORD CARRIERS
    • G06K9/00Methods or arrangements for reading or recognising printed or written characters or for recognising patterns, e.g. fingerprints
    • G06K9/00006Acquiring or recognising fingerprints or palmprints
    • G06K9/00067Preprocessing; Feature extraction (minutiae)

Description

1939- R. G. SAGEBEER 2,160,603
COMPAROMETER Filed Spt. 1, 1957 a Sheets-Shee t 1 May 30, 1939. R. G. SAGEBEER 2,160,503
COMPAROMETER Filed Sept. 1, 1951 8 Sheets-Sheet Richardlifla gebeev;
Show 1 8 Sheets-Sheet 4 R. G. SAGEBEER COMPAROMETER Fil'ed Sept. 1, 1937 May 30, 1939. I
M H c w w r I L.
y 1939- R. G. SAGEBEER 2,160,603
COMPAROMETER Fild Sept. 1, 1957 a Sheets-Sheet 5 s u Richard G-Sagebew;
y 1939- R. G. SAGEBEER 2,160,603
COMPAROMETER Filed ept. 1, 1937' 8 Sheets-Sheet 6 I Ribhard Gfiagebeer;
Q m k KM mm m. NR .mw mqwx. m m a a. fla fivqn, [*x RK Q N fix k$x km RR Al km .3 wrfi v a 3 m y .1939- R. G. SAGEBEER 2 ,160,603
COMPAROMETER Filed se tfl, 1957 8 Sheets-Sheet '7 I I .Fjlg-g l "g7 I 26' Il HIH 4 5 L l4 Richard as ebeer p I i f Y 55 14 I W y 3 1939- R. SAGEBEER 2,160,603
COMPAROMETER Filed Sept. 1, 1937 8 Sheets-Sheet s I i i 202 I I 2 ".llmu
' o E y- M Richard (ifiqyebeezg Patented May 30, 1939 UNITED STATES PATENT OFFICE COMPAROMETER Richard G. Sagcbeer, Wilmington, Del.
Application September 1, 1937, Serial No. 161,916
8 Claims. I (Cl. 88-44) This invention relates to the art of apparatus for comparing statistical data.
The general uses .of apparatus of the type to which the present invention relates, which will herein be called a comparometer; depend upon its property of computing quickly an index of closeness of fit between groups of numerical data.
The groups of data must have the same number of measures, i. e., certain variables must be chosen upon which to compare the groups. These variables are kept in the same order in all measurements, so that a certain value of the third variable, e. g., will always be compared with a certain other value of the same third variable. Each group of data gives a profile characteristic of the group.
In practice, each group of data may be compared in turn with every other group under consideration. This would be prohibitive without a means for making the comparisons very rapidly, as my comparometer does. When all the indices of closeness of fit have been determined, the groups may be divided into as many or as few classes as desired, depending upon the degree of closeness of fit which has been chosen as close enough for allowing two groups to be classified together. Each class will then contain only groups having similar profiles.
In some cases these classifications will be valu able at once, if the only desideratum is an unambiguous classification. In other cases the classes will have to be established, and then each class will have to be studied to determine its unique properties. When this has finallybeen done, and a new group appears which has the approximate profile of one of these recently established classes, that new group will presumably have the same, or nearly the same, properties as the class to which'it belongs.
A few examples of classifications where the classification itself is the primary consideration are: large files of photographs in a rogues gallery,
rge files of characteristic physical measurements such as are used to identify criminals, large files of census data, or any large body of groups of numbers which are to be classified and catalogued according to the pattern or profile of the numbers in a group.
A few examples of the second type of classification would be: educational measurements where a number 'of different measures are available for each individual student, vocational data, diag- 'nostic measures in the study of mental aberrations; m-eteorological'data, numerical indices of economic or financial trends, or any scientific measures which are complex and in which a number of numerical criteria have been applied to each one of a number of individual cases, and where the interdependence of the criteria is complex.
In the second type, of course, the establishment of the classification is only the first step. Thus, for instance, after the educational measurements have been classified, each class contains only students with very similar groupings of the values measured. The next step is to study each of these groups in order to determine the properties of each group. This is accomplished by discovering what properties, facilities, aptitudes, and limitations are held in common by all the 15 members of a certain classification and by no other students in the same way. Such properties thus become associated with the' Standard Profile of that particular classification. The discovery that some new individual possesses a 20 profile belonging in that classification will identify him as having similar properties, facilities, aptitudes and limitations.
It is important to note that the system of indices of closeness of fit existing between the members which have been so classified gives complete evidence necessary to determine the mathematical inter-relations between the classes. To be sure, the actual properties of the various classes must be determined by observing mem bers of each class. This observation, however, would give no evidence of the inter-relations existing between the classes. This inter-relation is determined from a mathematical analysis of the whole system of indices of closeness of fit, as obtained in the measurements made to classify the whole body of measurements.
Thus, the method not only classifies profiles, but furnishes complete mathematical material for determining the relationships between the 40 classes so derived.
It is therefore an object of this invention to compare related data so that the best set of factors for a given situation may be obtained.
It is a more specific object of this invention to provide apparatus by which a series of punched data cards may be compared with a given punched data card to determine the closeness of the comparison of the various cards with the first mentioned card.
It is a further object of this invention to provide a pneumatic method of comparing punched data cards.
It is also an object of this invention to provide a photometric method of comparing punched data cards.
With the above and other objects in view, which will be apparent'as the description proceeds, I have set forth my invention in the following I specification, and have illustrated it in the accompanying drawings in which:
Figure 1 is a perspective view of a pneumatic embodiment of my invention;
Fig. 2 is a transverse vertical section through Fig. 1;
Fig. 3 is a longitudinal vertical section on the line 3--3 of Fig. 2;
Fig. 4 is a plan view of Fig. 1 with the rails and profile card removed;
Fig. 5 is a plan view of a chart on which data is recorded to determine the location of the openings in the corresponding profile card;
Fig. 6 is a plan view of a profile card corresponding to the chart shown in Fig. 5;
Fig. 7 is a horizontal section on the line 1-1 of Fig. 3;
Fig. 8 is a vertical section on the line 8-8 of Fig. 7;
Fig. 9 is a plan view showing a modified form of construction having rectangular openings;
Fig. 10 is a transverse vertical section on the line I0I0 of Fig. 9 showing the block guides for the blocks of Fig. 9; i
Fig. 11 is a plan view of a profile card provided with diamond-shaped openings superposed on a construction having rectangular openings;
Fig. 12 is a plan view of a profile card having rectangular openings superposed on rails having spaced triangular openings;
Fig. 13 is a perspective View of a photometric embodiment of my invention;
Fig. 14 is a transverse vertical section on the line |4-|4 of Fig. 17;
Fig. 15 is a longitudinal vertical section on the line |5l5 of Fig. 14; i
Fig. 16 is a horizontal section on the line |6|6 of Fig. 15';
Fig. 17 is a plan view of the lower part of Fig. 13 showing the profile card and rails used therein;
Fig. 18 is a diagrammatic view of the electrical circuit containing the rheostats for the individual lamps;
Fig. 19 is a diagrammatic view of the electrical I circuit connecting the photoelectric cells;
Fig. 20 is a plan view of a rail containing a diamond-shaped opening for cooperation with a profile card containing spaced rectangular openmgs;
Fig. 21 is a longitudinal vertical section through Fig. 20;
Fig. 22 is a plan view of a rail containing spaced triangular openings for cooperation with a profile card containing single rectangular openings;
Fig. 23 is a longitudinal vertical section through Fig. 22;
Fig. 24 is a plan view, on an enlarged scale, of a portion of a rail containing a concaved diamond-shaped opening;
Fig. 25. is a plan view of a suitable punching machine; and
Figs. 26 and 27 are side and end elevations, re-' spectively, thereof.
Referring to the accompanying drawings, and particularly Figs. 1-8 thereof, I have disclosed a pneumatic embodiment of my invention.
This comprises a box 30 having three compartments, an upper compartment 3|, a middle compartment 32 and a lower compartment 33. Compartments 3| and 32 are separated by an airtight shelf 34, and compartments 32 and 33 are separated by an airtight shelf 35. The top compartment 3| is normally open at 36.
Air enters lower compartment 33 through a circular orifice 31, and passes into middle compartment 32 through a corresponding circular orifice 38, the same size as orifice 31, being drawn into compartment 32 through the suction created therein by an exhaust fan mounted on shelf 35, and arranged in proximity to a large orifice 4| in the rear wall 42 of compartment 32.
Air entering upper compartment 3|, as explained later, enters middle compartment 32 through a rectangular orifice 43 in shelf 34. The effective area of orifice 43 is controlled by a horizontal gate 44 provided with a triangular orifice 45. The gate 44 preferably comprises a fiat metal slide 46, terminating in a short vertical riser 41, for reciprocation in a. space in shelf 34'. Gate 44 is provided with any convenient handle having a scale 5| for registering with the front wall 52 of the box 30.
The relation of the pressures in the upper and lower compartments 3| and 33 is indicated by a manometer 53, of any suitable type.
At the sides of the top of the box 30, there are provided guides 54 and 55 to retain a plurality, in this case six, of movable rails 56 which, with profile cards described below, control the entrance of air into compartment 3|. As shown in Figs. 22 and 23, these rails preferably comprise strips of Wood 51, or any other suitable material, provided on their upper surface with a suitable protecting cover, such as the metal sheathing 56. In the embodiment shown in Figs. 1, 22 and 23, rails 56 are provided with a pair of triangularshaped openings 60 with the apices arranged towards one another, but spaced apart a predetermined distance.
As shown in dotted lines in Fig. 22 and in full lines in Fig. 23, the body portion 51 below the metal sheathing 58 is provided with rectangular openings 59,.which are appreciably larger than the triangular-shaped openings 60, in order that air and light passing through the triangularshaped openings may not be restricted by the body of the-rails.
The upper surfaces of rails 56 terminate inthe planeof shoulders 6| on guides 54 and 55, which shoulders provide a convenient support for profile cards 62 of any suitable material,-such as cardboard.
For use with the rails shown in Fig. 1, the profile cards 62 are preferably provided with a plurality, in this case six, of rectangular orifices 63 of equal size, which size corresponds to the specing between the apices of the triangular openings 60 in the rails 56. The orifices 63 in the particular profile card 62 shown in Fig. 6 correspond to the data points 64 on a predetermined chart 65, the centers of the orifices 63 coinciding with the corresponding data points .64.
When my embodiment, just described, is to be used, the operator will observe that the reading of the two columns of the manometer is even be- ,fore any card is inserted, because of the equal pressures in the upper and lower compartments 3| and 33. The operator then starts the fan 40,.
herent friction or by any suitable clamp or spring (not shown). The manometer reading now becomes unbalanced, because of the lower pressure in upper compartment 3| than in lower compartment 33, because air is prevented from going into compartment 3|. The operator now replaces the first profile card with another profile card without changing the position of the rails but with its scale mark 66 at the edge of the guide rail 54, and adjusts the gate 44' until the two columns of the manometer are even. The reading on the scale 5| on the handle 50 of the gate 44 is noted, and this is the degree of the closeness of fit of the orifices 63 in the second profile card with the orifices in the first profile card. The second card is then removed and the remaining cards are each tested in the same way. The scale readings for the various cards tell the operator which profile cards, and hence which sets of data, most closely match the data on the first profile card.
Each chart 65 and corresponding profile card 62 are provided with marginal marks 66. When a mark 66 coincides with the end of the guide rail 54, the middle of the profile card 62 and the mean value of the six variables will be at the middle of the space above the rails 56.
7 These variables will preferably be expressed in terms of the standard deviations of the respective variables, as is done in the well-known T-scale frequently used in educational measurements. Where six variables are to be used, they will be based on six measurements which are called raw scores. To convert these into T-scales, the following procedure is followed. The mean value of all scores is determined for each test. The devia tion of each score from the mean of all the scores is next determined. Eachdeviation is squared, and the sum of the squared deviations is divided by the number of scores to give the "mean square of the deviations. The square root of the latter is known as the Standard Deviation", and is the .unit to which statistical measures are usually reduced. Each raw score in turn is divided by the standard deviation 'of all scores on that test. Each quotient is the number of standard deviations, or fraction thereof, above or below the mean, of that individuals score. The corresponding T-scale score is derived by representing all scores of zero standard deviation as 50 on this scale. Each unit above or below 50 represents one-tenth of a standard deviation. Thus, a score on this scale of 80 is three standard deviations above the mean of all the scores,-and a score of 20 is three standard deviations below the mean. 1
With all scores, for all six tests, converted to this T-scale, the profile cards are prepared as follows:
The six scores for the first individual are averaged. The orifices are punched so that this average of the six scores is at the midpoints of the two guide rails 54 and 55 when the card is in its correct position. This correct position for the card is indicated when the marginal marks 66 coincide with the end of the guide rail 54.
Where desired, a small block or stop (not shown (may be provided at the rear end of the guiderail. Each card may then be inserted until it contacts with this stop without the necessity'of bringing the mark 66 into coincidence with the end of the guide rail 54.
As noted above, the center of each rectangular orifice in'a card will be a certain number of units (T-scale) above or below the mean of the six scores of the given data. The punching machine (desoribed later) is provided with a scale having one hundred equally-spaced divisions. Each of these divisions is one-tenth of a standard deviation, and the complete scale therefore is the so-called T-scale. It is necessary to callbrate the scale 5| on the gate handle 50 to read in similar units. In order to do this, a second scale must be provided, which is an exact duplicate of the T-scale (discussed later) on the punching machine. A profile card 62- with its perforations 63 is placed above the rails. Each rail 56 is adjusted until the rectangular orifice 63 in the card 62 above that rail is exactly closed by the solid portion of that rail lying between the apices of the two triangular openings 60 in the rail, as described above. In this position no air can pass through any of the orifices. The card 62 is now displaced toward the operator until the mark 65 is one unit from the end of shoulder 54, as indicated by the duplicate T-scale.
' The handle 50 is now adjusted with the fan 40 running until the manometer arms 53 are at equal levels. This position on the handle 50 is calibrated as one unit. The card 62 is next displaced two units, a new balance is obtained by moving handle 50, and the new position is calibrated as two units, and so on until the whole scale 5| has been correctly calibrated.
If any other card 62 is now inserted instead of the one to which the rails have been set, and if the manometer 53 is rebalanced by resetting the handle 50, the calibrated scale 5| will now read in tenths of a standard deviation the disparity between the first profile card and the one inserted last.
While in the foregoing embodiment of my invention, I have disclosed rails 56 having spaced triangular orifices 60, and profile cards having single rectangular orifices 63,.various other combinations of orifices may be utilized.
As indicated in Figs. 9 and 10, for example, the top of the upper compartment 3| may be provided with card guiderails 16 to guide profile cards18 having diamond-shaped orifices 80 as shown in Fig. 11. In this embodiment .of my invention, the top 15 also has a'plurality of block guides 8| to support slidable blocks 82 which leave rectangular orifices "l'l in top 15 on either side of the slidable blocks.
' As shown in Fig. 11, the diamond-shaped orifices 80 may be made of varying widths when different weightings of the variables are desired, but the space between the ends of the corresponding rectangular orifices 11 must be the same as the length of the corresponding diamond-shaped orifice 80 in the profile card 18 so that the blocks 82 will close the diamond-shaped orifices in the card whenthey are exactly beneath them. Of course, where all of the variables are to have the same weighting, the diamond-shaped orifice would have equal widths.
As shown in Figs. and 21 each rail 86 may be provided with a single diamond-shaped orifice 81 having an enlarged rectangular opening. 88 below the top sheathing. "With this type rail, the corresponding profile card will have pairs of spaced rectangular-shaped orifices, as shown in Fig. 17.
Also, where desired, a rail 96 may have 2. diamond-shaped orifice 91 with concaved sides 98 as shown in Fig, 24. The corresponding profile card is shown in Fig, 17.
For the sake of compactness, the dimensions of the rails 56 and of the chamber 3| beneath the rails bear definite relations to the length of the rectangular orifices .63 used. The altitude of a triangular orifice 60 is preferably equal to the length of a rectangular orifice 63. The length of a diamond-shaped orifice 81,91 or I49 is preferably twice the length ofthe rectangular orifice with which it is used. The'maximum length of the scale used should always be at least twice the length of the rectangular orifice.
When one rectangular orifice 63 in the profile card is used with two triangular orifices 60 in the rail, as in Fig. l, the length of the chamber 3I beneath the rails 56 must be at least five times the length of one rectangular orifice 63, and the rail 56 in this case must be at least seven times'as long as the rectangular orifice 63. When these relationships are reversed, so that there are spaced triangular orifices in the profile card and a single rectangular orifice in each rail, the rail must be at least five times the length of the rectangular orifice, and the chamber 3| must be at least three times the length of the rectangular orifice.
When two-spaced rectangular orifices I19 in the profile card are used with one diamondshaped orifice I49 in a rail I46, as in Fig. 17, regardless of whether the diamond-shaped orifices are concave or straight, the chamber must be at least six times the length of a single rectangular orifice I19, and the rail I46 must be at least eight times the length of a single rectangular orifice I 19.. When these relationships are reversed, so that there are spaced rectangular orifices in each, rail and single diamondshaped orifices in each profile card, the rail must be at least eight times the length of a single profile card 18 are used with slidable blocks 82 in the top 15 of a compartment 3|, as in Figs. 9-11, the blocks 82 must be of. the same length as the length of the diamond-shaped orifices 80, and the block guides 8|, which are the length of the interior of compartment 3I, must be at least three times the length of the blocks 82.
For practical purposes, I have found that coefficients of closeness of fit with less than one per cent error are satisfactory. This" is obtained byusing a rectangle three inches long, with a six-inch scale divided into one hunded equal divisions. Smaller rectangles could be used where this degree of accuracy is not necessary. Similarly, where still greater accuracy is desired, this may be accomplished in well-known ways, such as by using veTnierspr reading lenses. The preferred width of rail, when six are used, is two inches. If the number of variables is more than six, each rail may be correspondingly narrower without loss of accuracy in the final coeillcient of closeness of fit.
In the embodiment of my invention shown in Figs. 13-19, I have ,disclosed a photometric,embodiment of. my invention. This comprises a box I30, having an upper section I3I and a lower section I32, which are adapted to fit together by the engagement of a projecting rib I33 on section I3I between guide rails I34 and I35 on section [32. The rib I33 terminates in a plane above the shoulders I36, which are formed on the ribs I34 and I35 to provide a space I40 for a profile card as discussed below.-
The upper section I3I comprises any desired number of light-projection compartments I31, in this case six of these compartments being 'shown. Each of these compartments is formed by parallel walls I38, but is open at both its top and bottom, and the walls of adjacent compartments, while spaced apart at .their upper ends, come to a common line at their lower ends in order to bring the open bottom portions of the compartments closer together for registering with a pluralityof rails I46, corresponding to that shown in Fig. 20, each of which is provided with a diamond-shaped orifice I49.
I provide a slidable lamp block I41 for each of the compartments I31, each said lamp block having a lamp I48 mounted therein in such a way as to throw the light downwardly into the corresponding compartment I31. The open upper ends of the various compartments I31 may be closed by a suitable cover (not shown).
The lower section I32 comprises a large com- I partment II and two smaller compartments I52 and I53. Compartment I5I is separated from compartments I52 and I53 by a sloping wall I54 extending from the top near one side of this section to the bottom thereof, and compartments I52 and I53 are separated by a light-proof. shelf I55, which is provided with a suitable triangular orifice I 56 that may be closed or uncovered by the operation of a gate I51 of the same type as gate 44 discussed above, except that gate I51 has no orifice therethrough. Gate I51 is provided with a handle I58 carrying a scale I60. Gate I51 operates in a space I6I provided in shelf I 55. Compartment I52 is provided with a lamp I62 mounted on the front wall of that compartment. Its rear wall comprises a downwardly and in Wardly sloping mirror I63 which is adapted to reflect light from lamp I62 through the triangular orifice I56 in shelf I55 when gate I51 uncovers orifice I56. This is reflected on to the white floor I64 of compartment I53 and back to a photelectric cell I65 mounted on the under side of shelf I55.
Light from lamps I48 mounted in the compartments I31 of upper section I3I passes through the corresponding orifices I49in rails I46 into main compartment I5I, and is received on a sloping rectangular mat I66, which is preferably white, which reflects the light backto a photoelectric cell I 61 mounted in the upper corner of compartment .I5I substantially parallel to sloping rectangular mat I66.
Photoelectric cells I 65 and I61 are connected in series in a circuit I68 with a switch I69 by means of which photoelectric cell I65 may be short cir'cuited, as shown in Fig. 19, with a millivoltmeter I and a rheostat I provided with an adjusting knob I12. The photoelectric cells I65 and I61 are of opposite polarity so that with equal illumination upon them no voltage will be impressed on millivoltmeter I10. The variable rheostat "I is used to protect the millivoltmeter until the blanace has been approximated.
The photoelectric cells I65 and I61 described here are preferably the so-called dry copper oxide type, which generate an electromotive force when illuminated. Any other type of photoelectric cell, such as the potassium cell which liberates charges into a vacuum when illuminated, could be used instead by providing the necessary equipment, such as batteries and amplifiers if. necessary.
Each of the lamp blocks I41 is provided with a small rheostat I 13 with an adjusting knob I14 to control the illumination furnished-by each lamp I46. All of the lamps I46 and lamp I62 are connected in parallelin a circuit I15. Lamp I62 is controlled by a'rheostat I16 provided with an adjusting knob I11. 7 These rheostats' I13 are used to allow the various light rays from the lamps I48 to be adjusted to exact equality in cases where the lamps may not be of exactly equal intensities. These rheo-' previously calibrated by inserting a card I18,
adjusting rails I46, then displacing the card small 3 unit displacements and marking unit calibrations,
to correspond to these unit displacements, on the handle I58, as described above for the calibration of the pneumatic embodiment. 4
When it is desired to operate my photometric I embodiment, a profile card I18, provided with rectangular openings I19, as shown in Fig. 17, is
inserted in the space I40 left between upper and 'lower sections I3I and.I32, and rails I46 are adjusted until no light reaches photoelectric cell I61. Lamp blocks I41 containing the lamps I48 are-then adjusted sothat each lamp H8 is di-- rectly above the diamond-shaped orifice. I49 in the corresponding rail I46 below the openvlower end of that compartment 131. I The first profile card I18 is replaced by a second one, and the gate I51 is adjusted until the millivoltmeter I10 reads zero. The reading for this particular profile card is indicated by the reading on the scale I60, of handle I58, which is just visible beyond the front wall of section I32.
The second profile card is replaced by a third profile card and so forth, adjusting the gate I51 ca ch time until the millivoltmeter l10 reads zero, andnoting the reading on the scale I60. These scale readings tell the operator which profile cards, and hence which sets of data, most closely match the data-onthe first profile card.
The preferred height for the lower section I32 of my photometric embodiment is thirty-two inches.- If this size is decreased-it becomes diillcult to converge the beams of light from the various rails I46 at the center of the floor I66 of the section.
The various orifices 63, 80 and I19 in profile cards 62, 18 and I18, respectively, may be made therein in any desired way, as by the use of any suitable punching machine 200.
The latter comprisesfour essential parts: A table 20I, a T-square 202 with a guide rail 203 for aligning a profile card 204, an arm 205 with knives 208 andguides 209, and a T-scale 2I2.
The table 20I has rigidly attached to it the guide rail 203 for the T-square 202, and also two blocks 206 carrying a bolt 201 upon which the arm 205 carrying the. knives 208 is pivoted. The T -square has a grooved head 2I0 for sliding along theeguiderail 203 and a blade 2 to align the card. "The exact vertical position at which the orifices ,will be punched'is determined by the T-scale 2I2, the bottom edge of the card being aligned with the desired score on this scale before making each orifice. The two guides 209 prevent anylateral displacement 'of arm 205 during a perforating operation.
In use, the head 2I0 of the T-square202 is set for the first position 2, the card 204 is placed 6 against the right edge of the blade 2I I, and the lower edge 'of the card 204 is aligned with the first score on the T-scale 2I2. The arm' 205 is depressed and the orifices made. The T-squa'r 02 is next shifted laterally to the second posi o'n 2I4 on the guide rail 203, thecard 204 is adjusted again to touch the blade 2I I and the second score on the T -scale 2I2, ,and the orificesmade, and so on.
All of the profile cards are provided with marginal marks 66which must coincide with the end. of the guide rail in order to center the mean value for the data recorded 'on that particular profile card in the space above the rails. I
As disclosed above, the shape of the orifices in the profile card must bear a suitable relation to the shape of the orifices in the rails used in conjunction with that profile card-but various combinations of these orifices may be used. As I shown in Figs. 1 and 12, for example, single rectangular orifices 63 in profile card 62 may be used with spaced triangular-orifices 60 in. rails 56. As shown in Fig. 1'1, spaced rectangular orifices I19 in profile card I18 may be used with single diamond-shaped orifices I49 in rails I46.
Various other combinations, may alsobe used.- For example, each of these arrangements may be reversed, as by providing a profile card for Fig. 1 with spaced triangular-orifices which cooperate with single rectangular orifices in the rails. r Similarly, the diamond-shaped orifices in either the. profile cards or the rails may. be of the concave side type 91 shown in Fig. 24,'and these concave side diamonds may replace the straight side ones,
shaped openings in the rails.
' As also shown in Figs. 9, 10 and 11, the rails provided withv orifices may'be replaced with top members 15 provided with block guides 8 I to re- 50 oeive slidable blocks 82 that leave rectangular orifices 11 on either side thereof, and profile cards 18 used in conjunction with the orifices 11 may have single diamond-shaped orifices 80.
profile card 204 under the vknives 208. Guide rail 203 is provided with marlns'2'I4 to indicate the proper displacement of the card 204 for punching each pair of orifices 2I5. For the embodiments represented in Figs. 1', 6, 1-2 and- 1'1, guide rail 203 would have six equally spaced marks 2 for the six punching positions required. The arm 205, which is of any desired material, suchas wood, carries knives 200 designed to cut the desired shape of orifice 2I5 in theprofilecards 204. In the drawings, two rectangular knives 200 are When the handle 2I3 of arm 205 is depressed, the knives 200 are driven through the It is always necessary, however, that the ori- 1 fices and solid portions ofthe corresponding rails and profile cards be such that the solid portions betweenspaced orifices in one member can just completely cover the orifices in the other member,
and that one set of orifices have similar sloping v sides since my method operates on the principle of least squares, and the best measure of fit occurs when the sum of the squaresisa minimum.
In all cases, the center of each orifice in a profile-card, regardless of the shape of the orifice, q-
or the center of thesolid portion between two orifices, must coincide with the corresponding data point on the data chart. Instead of the triangular or diamond-shaped orifices, any other type of rail may be usedifit has the sameproperty. as is possessed by these orifices, namely of permitting a flow which in-' creases in proportion to the square of the exposed height of the trianglesame property could be provided in the photometric embodiment, for
example, by using a so-called optical wedge instead of triangles or diamonds. Each optical wedge would form a window opening through a any suitable material, such as heavy paper, could be used. This would be carried on a roller, and
unwound through the comparometer and wound simultaneously onto a second roller with the aid of any suitable indexing device. The successive profiles would be perforated through this strip of paper at intervals instead of being punched onto individual cards.
card with no mark 66 is placed above the rails,
and if the rails have previously been set for any pattern whatever having its mean or average value at-the exact center of the guide rails, the
correct position for the mark 66 on the unmarked card will be found by moving card backward or forward until the indicator (manometer or millivoltmeter) is at a. minimum reading. In this po-- sition the mark will be correctly placed on the point of the card which coincides with the end of the guide rail.
Weighting of the various variables can in any case be accomplished by having various rails for the various weightings, and by substituting new rails having predetermined different sized orifices to give new weighting values. The weighting is T proportional to the base of the triangular orifice,
its use in connection with the prediction of stock market movements based on the way the market has acted at a previous time, when most of these factors were the same. For this purpose, if the cards were to be punched for these factors, this might beas follows:
(0) Volume of sales (b) Steel production K (0) Electric power consumption s (d) Freight car loadings (e) Bank clearings (f) Brokers loans on the guide card, and fromone end to the other of the card there would be in effect a graduated scale which would be different for each of these factors. Consequently, when the card was punched six times, once for each of these factors,
at the position on the card corresponding to.
the magnitude of the given factor, the first card would record the given positions for each of these be foretold by finding out what happened in the bases for classification. The classification might Y (a) Coefficient of diversity of interests next week based on the Dow-Jones averages or a given stock or something else that might be desired. By making out one of these cards for each week in the year, and repeating this for two years, there would be obtained one hundred four cards. Of course, this might be carried out for any desired length of time. If this had been done for the last two years, then the probable action of the market for the current-week might preceding-one hundred three weeks where these factors were substantially the same by comparing the card for the last week (which, of course, sets the-stage for the current week) with each of the otherone hundred three cards. "When this was carried out by the use of the present apparatus, it might be found thatv there were two or three cards out of the one hundred three which fairly closely approximated the arrange-' ment of these six factors on last week's card.
By looking at the indications of what the stock market did when these factors obtained previously, it would be reasonable to presume that, with the same set of factors for last week, the same change in the market would take place for the current week.
The procedure is the same for all of the examples mentioned. In classifying photographs, certain standard axes would be chosen. The distances of marginal points on the profile from one of these-axes would be measured at regular intervals, and the measurements transferred to a card marked. with that individuals identification.
- If the number of individuals is not too large, each card would be matched against each of the other cards in-tum. If the "population is excessive- .ly large, the same results will be obtained if-a random sampling is madefrom the population and if the sampling is then subjected to analysis by this method. Statistical methods ofler principles which can be applied to determine how large a sampling must be made to give any desired degree of accuracy in this procedure.
In the census analysis there might be many be made upon five measures or criteria such as:
(a) Number in family (b) Amount of property owned (0) Density of population at point of residence (d) Total income of family (e) Type of occupation Tchosenmightbe: Each of these factors would have a givencolumn (0) Mental age'or intelligence quotient (b) Numerical aptitude (c) Verbal aptitude (d) Linguistic aptitude (e) Space-conceptaptitude (f) Finger-dexterity measurement (h) Coemcient of independence of eifort On such a" basis, the analysis would give' certain classifications of patterns as well as the technical inter-relations between these classifications. I
The actual properties of these various classes would then be determined from study of the relative success in various pursuits of these persons in later life. Study of measurements associated with mental cases would be treated in a similar manner. I
In meteorology, the variables chosen would be the meteorological measurements that are most significant in prognosis of weather. Some of these trends, such as proximity of an existing low area to a characteristic and typical lowpressure track (lows move across the continent from west to east or up the coast along certain preferred tracks), would be subjected to separate classification'or analysis into sub-classes. Finally, typical patterns of meteorological conditions would be determined by matching each individual pattern, say for one particular day, against all the other patterns. It would then be possible to discover what kind of weather would be most likely to follow afterthe appearance of any of these typical patterns. By means of the technical analysis of the inter-relations existing between these patterns, it would be possible to infer much about some pattern of an eccentric or unique character, even though it would not be close enough to any of the "standard patterns already set up. After these inferences have been checked against the actual weather ensuing, this unique pattern would be included in the standard patterns. It is interesting to note that there would-be no difiiculty in making a place for it in the classification. No rearrangement would be necessary. The nature of the system is such that each new pattern, as it presents itself for the first time, can be classified at once, and can be located at once in the files when the need arises. From the above explanation it will be apparent thatI have developed a method, and two preferred embodiments for carrying out that method, for quickly relating a series of variables to a plurality of other series of variables to determine the closeness of fit of these different series of variables.
It will also be apparent that this method, and these preferred embodiments, may be used to determine the relation between various series of variables for the proper classification of series of variables and for the prediction of the consequences of the occurrence of a given set of variables based on the consequences which followed a previous occurrence of that set of variables.
As many apparently widely different embodiments of this'invention mg be'made without departing from the spirit thereof, it is to be understood that I do not limit myself to the foregoing embodiments or description except as indicated in the following patent claims:
I claim:
1. In a photometric device for comparing predetermined sets of data, the combination of a housing having two sets of compartments of two compartments each with a light source in'one compartment of each set, an opening between the compartments of each set to permit light to pass from one to the other, a plurality of perforated rails arranged side by side to fill the space between the compartments of one set of compartments, a plurality of differently perforated profile cards, including a standard profile card, adapted to be placed one at a time in cooperative relation with the rails, means supporting said rails and card for relative adjustment to positions 'such that the rail and card openings are just out of registration, whereby no light passes from one compartment to the other, a millivoltmeter, a light sensitive cell in the compartment of each set not containing the light source, said millivoltmeter and cells being connected in an opposed series circuit for indicating the relative amounts of light in said compartments, and an adjustable shutter between the compartments of the second set for varying the size of the opening therebetween and restoring the millivoltmeter to its original reading when the standard profile card is replaced by another card with a resulting difference in registration of the rail and card openings and a change in the relative amounts of light in the two compartments.
2. The photometric device of claim 1, in which the adjustable shutter has a scale for indicating the amount it is moved.
3. The photometric device of claim 1, in which the light source in the first set of compartments comprises a separate light source for 'co-action with each rail, and means are provided for preventing light from each light source from reaching all other rails than the one with which it co-acts.
' 4. The photometric device of claim 1, in which the light source-containing compartment of the first set is divided into a plurality of sub-compartments, corresponding in number and align.- ment with the perforated rails, with a separate light source in each sub-compartment for transmitting light through the corresponding rail when the perforations in the rail and profile card are in registration.
5. The photometric device of claim 1, in which the light source-containing compartment of the first set is divided into a plurality of sub-compartments, corresponding in number and alignment with the perforated rails, with a separate light source in each sub-compartment for transmitting light through the corresponding rail when the perforations in the rail and profile card are in registration, and means for independently adjusting the intensities of the separate light sources.
6. The photometric device of claim 1, in which the openings in the rails and cards. are such that the openings in one of these members are spaced triangular openings with the apices turned toward one another, and the openings in the other member are single rectangular openings, and in which the multiple openings are spaced apart a distance equal to the length of the single openings.
7. The photometric device of claim 1, in which the openings in the rails and cards are such that the openings in one of these members are spaced rectangular openings and the openings in the other member are single diamond shaped open-i ings, and in which the multiple openings are spaced apart a distance equal to the length of the single openings.
8. A method for comparing predetermined sets of data to determine their closeness of fit to a given set of data, which comprises arranging a series of orifices in parallel relation in accordancewith the given set of data in such a way that .their relative positions determine the magnitudes of the items of data, arranging another set of orifices in relation to the first set so that they are just out of registration and the passage of flux through said first-mentioned orifices is blocked, substituting a different set of orifices arranged according to a different set of data for said first-mentioned set to permit the passage of an amount of flux, depending on the difierent arrangement of the orifices, adjusting a master orifice which is independent of said other orifices to change the amount of fiux passing through it by an amount sufiicient to compensate for the differences in the orifices until a zero reading is obtained on an indicating means which shows the relative magnitudes'of the flux passing through the first-mentioned orifices and said master orifice, noting the amount said master orifice was adjusted, and successively substituting other sets amount said master orifice was adjusted each time, and comparing the amounts of said adjustments to determine the amounts of flux which 10 passed through the various sets of orifices.
RICHARD G. SAGEBEER.
US161916A 1937-09-01 1937-09-01 Comparometer Expired - Lifetime US2160603A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US161916A US2160603A (en) 1937-09-01 1937-09-01 Comparometer

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
US161916A US2160603A (en) 1937-09-01 1937-09-01 Comparometer

Publications (1)

Publication Number Publication Date
US2160603A true US2160603A (en) 1939-05-30

Family

ID=22583354

Family Applications (1)

Application Number Title Priority Date Filing Date
US161916A Expired - Lifetime US2160603A (en) 1937-09-01 1937-09-01 Comparometer

Country Status (1)

Country Link
US (1) US2160603A (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US2730037A (en) * 1951-10-12 1956-01-10 Frederick K Stark Selector mechanism
US2883558A (en) * 1953-10-28 1959-04-21 Contraves Ag Function generating apparatus
US2950799A (en) * 1952-02-07 1960-08-30 Alan Foster Apparatus for identifying paper money, or the like, as genuine, and for making change or the like
US3023318A (en) * 1960-01-27 1962-02-27 John H Jones Optical switching
US3235741A (en) * 1961-04-24 1966-02-15 Invac Corp Switch

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US2730037A (en) * 1951-10-12 1956-01-10 Frederick K Stark Selector mechanism
US2950799A (en) * 1952-02-07 1960-08-30 Alan Foster Apparatus for identifying paper money, or the like, as genuine, and for making change or the like
US2883558A (en) * 1953-10-28 1959-04-21 Contraves Ag Function generating apparatus
US3023318A (en) * 1960-01-27 1962-02-27 John H Jones Optical switching
US3235741A (en) * 1961-04-24 1966-02-15 Invac Corp Switch

Similar Documents

Publication Publication Date Title
Duncan A socioeconomic index for all occupations
Galton Fingerprint directories
US2160603A (en) Comparometer
Prokić et al. Detecting shibboleths
McCroskey et al. Attitude intensity and the neutral point on semantic differential scales
US4309111A (en) Visual coin grader
Kefauver et al. The Secondary-School Population
Mulligan et al. Adaptors, innovators and the Kirton adaption-innovation inventory
Chahbazi Use of projective tests in predicting college achievement
Frear Photoelectric apparatus for measuring leaf areas
Richards Jr et al. Regional differences in junior colleges
Gavett A first course in statistical method
Bean et al. Size, functional specialization, and the classification of cities
Coffey A quantitative measure of bias in reporting of political news
Wright et al. Performance tests with non-circular coins
Peterson Testing deaf children with Kohs block designs
US3522655A (en) Apparatus and method for determining the correlation coefficient of two graphical traces
Scheips A pattern analysis of job requirement factors for a sample of jobs
Schufletowski et al. Value Structure: Another Dimension to Attitude Measurement?
US2122440A (en) Scale duplicator
Stricker Measuring Social Status with Occupational Information: Some Useful Procedures.
Sabloff The decision process of a professionalized legislature: A cognitive anthropology approach
Show Planning, constructing, and operating forest-fire lookout systems in California
US2036439A (en) Apparatus for teaching double entry bookkeeping
Stene Newspapers in the Presidential Campaign